athena-reader

KO | EN

gitlite — search

#hacktoberfest #golang #android #machine-learning #rust #deep-learning #docker #nodejs #macos #cli #kubernetes #vue #pytorch #ios

(★ 17)

A simple speed reading application that supports .txt, .pdf, and image text extraction

athena-reader 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Athena Reader

Athena Reader is a native desktop speed-reading tool:

- Paste text or a screenshot from your clipboard
- Run OCR locally (offline after models are cached)
- Stream the extracted text at a target WPM using an ORP-style centered word display

This repository is a Rust workspace with:

- `athena-core`: OCR + text normalization/tokenization + reading session logic
- `athena-app`: egui/eframe GUI application

## Capabilities at a glance

- Import files: image (`png/jpg/jpeg/bmp/tif/tiff/webp`), PDF, EPUB, and `.txt`
- Paste text or image from clipboard (**Ctrl/Cmd+V**)
- Non-blocking background processing for import/OCR/extraction
- ORP-style centered reading display with configurable WPM, chunk size, font size, and theme
- Read-only preview with separate **Edit Text** window (double-click preview)
- **Live View**: a "+" button opens a separate window showing all text with the currently-streaming word highlighted and auto-scrolling — tracks the stream in real time, even when the main window has focus
- Editor-open pauses playback; editor-save rebuilds playback text and remaps reading position
- Local persistence of user settings and paused reading position across app restarts

## How it works (User flow)

1. Launch the app.
2. Bring text into the app:
   - **Import**: pick an image file, a PDF, an EPUB, or a `.txt` file.
   - **Paste**: paste text or an image from the system clipboard (button or **Ctrl/Cmd+V**).
3. Processing runs in a background thread (UI stays responsive).
   - OCR from images typically takes a few seconds depending on the image size/quality and your machine.
4. The bottom preview shows the first **25 words** of extracted text.
5. Double-click the preview to open **Edit Text** in a separate window.
   - If playback is running, opening the editor pauses it.
   - **Save** rebuilds the reading session from edited text and keeps the closest logical reading position; if the old position is no longer valid, it resets to the start.
6. The reading view shows the first word/chunk immediately; press **Play** (or **Space**) to begin streaming.
7. While paused, navigation/edit saves update the cached reading snapshot; reopening the app restores that paused text/position.

## Controls / Shortcuts

- Paste clipboard text/image: **Ctrl/Cmd+V**
- Play/Pause: **Space**
- Step back/forward by 5 words: **Left / Right arrow**
- Restart (resets to first word and pauses): **R**

## Import behavior and limits

| Input type | Processing path | Notes |
| --- | --- | --- |
| Image files | OCR (`oar-ocr`) | Runs locally once models are available. |
| Clipboard image | OCR (`oar-ocr`) | Clipboard must contain image data. |
| PDF | `pdf-extract` embedded-text extraction | Scanned/image-only PDFs usually return empty text and show an error. |
| EPUB | `epub-stream` chapter extraction | Extracts plain text from spine chapters in reading order. |
| `.txt` | UTF-8 lossy decode (`String::from_utf8_lossy`) | Invalid UTF-8 bytes are replaced with replacement characters. |
| Clipboard text | Direct text load | Whitespace is trimmed; if both text and image exist, text is preferred. |

## Persistence behavior

- **Settings**: WPM, chunk size, font size, and theme are saved to `settings.json` in the app config directory.
- **Paused reading cache**: text + current token index are saved to `reading_cache.json` in the app config directory.
- On startup, the app restores a valid paused session automatically; invalid/out-of-range cached indices reset to index `0`.

## UI Components (Quick Guide)

- **Top controls**
  - **Import**: select an image, PDF, EPUB, or text file.
  - **Paste**: loads clipboard text directly when available; otherwise falls back to image OCR.
  - **Play / Pause / Prev / Next / Restart**: playback controls.
  - Sliders:
    - **WPM**: 100–900
    - **Chunk**: 1–5 words per step
    - **Font**: 18–200
  - **Theme**: Light / Dark / High Contrast
- **Preview panel**
  - Shows the first 25 words of extracted text.
  - Double-click to open **Edit Text**; **Save** rebuilds the reading session from the edited text (paused).
  - **"+" button** (bottom-right): opens the **Live View** window.
- **Live View window**
  - Separate OS window displaying all tokens as flowing text.
  - The current word (or chunk) is highlighted with a theme-aware color and underline.
  - Auto-scrolls to keep the highlighted word in the top 1/3 of the viewport.
  - Adjustable font size slider (18–200).
  - Mirrors all playback actions (play, pause, seek) from the main window in real time.
  - Opening the Live View pauses playback; closing it does not change playback state.
- **Bottom-right status**
  - Shows the current status plus a processing timer/spinner while OCR/PDF extraction is running.

## OCR backend and models

The default OCR backend is:

- `oar-ocr` (ONNX models executed via ONNX Runtime)

Default model artifacts are downloaded from Hugging Face (MIT licensed):

- Repo: `GetcharZp/go-ocr`
- Files:
  - `paddle_weights/det.onnx`
  - `paddle_weights/rec.onnx`
  - `paddle_weights/dict.txt`

Models are cached locally in the OS cache directory (via `directories::ProjectDirs`), under:

`~/.cache/athena-reader/ocrs/` (Linux default)

`%LOCALAPPDATA%\\athena\\reader\\cache\\ocrs\\` (Windows default)

## Environment variables

### Model paths (local files)

- `ATHENA_OCR_DETECTION_MODEL`
- `ATHENA_OCR_RECOGNITION_MODEL`
- `ATHENA_OCR_DICT_PATH`

### Model downloads (override URLs and optional integrity pins)

- `ATHENA_OCR_DETECTION_URL` / `ATHENA_OCR_DETECTION_SHA256`
- `ATHENA_OCR_RECOGNITION_URL` / `ATHENA_OCR_RECOGNITION_SHA256`
- `ATHENA_OCR_DICT_URL` / `ATHENA_OCR_DICT_SHA256`

### Cache directory + download toggle

- `ATHENA_OCR_MODEL_CACHE_DIR` (override the directory where models are stored)
- `ATHENA_OCR_DISABLE_DOWNLOAD` (set to `1`, `true`, `yes`, or `on` to prevent downloading)

## Build and run

### Prerequisites

- Rust toolchain (stable) + Cargo

> Note: `oar-ocr` uses ONNX Runtime and may download native runtime binaries depending on its configuration/features.

### Run the app

```bash
cargo run -p athena-app
```

### Build

```bash
cargo build --release -p athena-app
```

## Testing / Linting

```bash
cargo fmt
cargo clippy --workspace -- -D warnings
cargo test --workspace
```

athena-reader

athena-reader 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

athena-reader

athena-reader 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README